momentuHMM: R package for generalized hidden Markov models of animal movement

摘要

Discrete-time hidden Markov models (HMMs) have become an immensely populartool for inferring latent animal behaviors from telemetry data. Here weintroduce an open-source R package, momentuHMM, that addresses many of thedeficiencies in existing HMM software. Features include: 1) data pre-processingand visualization; 2) user-specified probability distributions for an unlimitednumber of data streams and latent behavior states; 3) biased and correlatedrandom walk movement models, including "activity centers" associated withattractive or repulsive forces; 4) user-specified design matrices andconstraints for covariate modelling of parameters using formulas familiar tomost R users; 5) multiple imputation methods that account for measurement errorand temporally-irregular or missing data; 6) seamless integration ofspatio-temporal covariate raster data; 7) cosinor and spline models forcyclical and other complicated patterns; 8) model checking and selection; and9) simulation. momentuHMM considerably extends the capabilities of existing HMMsoftware while accounting for common challenges associated with telemeterydata. It therefore facilitates more realistic hypothesis-driven animal movementanalyses that have hitherto been largely inaccessible to non-statisticians.While motivated by telemetry data, the package can be used for analyzing anytype of data that is amenable to HMMs. Practitioners interested in additionalfeatures are encouraged to contact the authors.